{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:ALIXB4WVUDKIKTEPLQKVN2JEEC","short_pith_number":"pith:ALIXB4WV","schema_version":"1.0","canonical_sha256":"02d170f2d5a0d4854c8f5c1556e92420995f1a258f63b4598d84a09d66131812","source":{"kind":"arxiv","id":"2402.14146","version":3},"attestation_state":"computed","paper":{"title":"Dynamic Multi-Reward Weighting for Multi-Style Controllable Generation","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Dongyeop Kang, Karin de Langis, Ryan Koo","submitted_at":"2024-02-21T22:02:37Z","abstract_excerpt":"Textual style expresses a diverse set of information, including interpersonal dynamics (e.g., formality) and the author's emotions or attitudes (e.g., disgust). An open question is how language models can be explicitly controlled so that they weave together target styles when generating text: for example, to produce text that is both negative and non-toxic. One approach to such controlled generation is multi-objective reinforcement learning (RL), but how best to combine multiple objectives in a reward function is an open question. In this paper, we investigate various formulations of multi-sty"},"verification_status":{"content_addressed":true,"pith_receipt":true,"author_attested":false,"weak_author_claims":0,"strong_author_claims":0,"externally_anchored":false,"storage_verified":false,"citation_signatures":0,"replication_records":0,"graph_snapshot":true,"references_resolved":false,"formal_links_present":false},"canonical_record":{"source":{"id":"2402.14146","kind":"arxiv","version":3},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2024-02-21T22:02:37Z","cross_cats_sorted":[],"title_canon_sha256":"5e689ad6483a947a4157149bdc8d7fdecc14546979beb4526d05ec4d2d194668","abstract_canon_sha256":"71b224f3dcc3875dd70f2a21f2f85735920fbbe066a0bba42acbadcf89bf2d08"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:39:39.816844Z","signature_b64":"YE1Z8FB5CABvC+YANVUnpXRDCcgSgLMSjDHOPnsGyqYNNyrqVz/WaUWiQ82HlDNuGoiqMEgtnW4f/1moHoYLBw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"02d170f2d5a0d4854c8f5c1556e92420995f1a258f63b4598d84a09d66131812","last_reissued_at":"2026-07-05T11:39:39.816370Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:39:39.816370Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Dynamic Multi-Reward Weighting for Multi-Style Controllable Generation","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Dongyeop Kang, Karin de Langis, Ryan Koo","submitted_at":"2024-02-21T22:02:37Z","abstract_excerpt":"Textual style expresses a diverse set of information, including interpersonal dynamics (e.g., formality) and the author's emotions or attitudes (e.g., disgust). An open question is how language models can be explicitly controlled so that they weave together target styles when generating text: for example, to produce text that is both negative and non-toxic. One approach to such controlled generation is multi-objective reinforcement learning (RL), but how best to combine multiple objectives in a reward function is an open question. In this paper, we investigate various formulations of multi-sty"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2402.14146","kind":"arxiv","version":3},"verdict":{"id":null,"model_set":{},"created_at":null,"strongest_claim":"","one_line_summary":"","pipeline_version":null,"weakest_assumption":"","pith_extraction_headline":""},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2402.14146/integrity.json","findings":[],"available":true,"detectors_run":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938"},"references":{"count":0,"sample":[],"resolved_work":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57","internal_anchors":0},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"author_claims":{"count":0,"strong_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"builder_version":"pith-number-builder-2026-05-17-v1"},"aliases":[{"alias_kind":"arxiv","alias_value":"2402.14146","created_at":"2026-07-05T11:39:39.816427+00:00"},{"alias_kind":"arxiv_version","alias_value":"2402.14146v3","created_at":"2026-07-05T11:39:39.816427+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2402.14146","created_at":"2026-07-05T11:39:39.816427+00:00"},{"alias_kind":"pith_short_12","alias_value":"ALIXB4WVUDKI","created_at":"2026-07-05T11:39:39.816427+00:00"},{"alias_kind":"pith_short_16","alias_value":"ALIXB4WVUDKIKTEP","created_at":"2026-07-05T11:39:39.816427+00:00"},{"alias_kind":"pith_short_8","alias_value":"ALIXB4WV","created_at":"2026-07-05T11:39:39.816427+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2605.14350","citing_title":"Distributionally Robust Multi-Task Reinforcement Learning via Adaptive Task Sampling","ref_index":270,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/ALIXB4WVUDKIKTEPLQKVN2JEEC","json":"https://pith.science/pith/ALIXB4WVUDKIKTEPLQKVN2JEEC.json","graph_json":"https://pith.science/api/pith-number/ALIXB4WVUDKIKTEPLQKVN2JEEC/graph.json","events_json":"https://pith.science/api/pith-number/ALIXB4WVUDKIKTEPLQKVN2JEEC/events.json","paper":"https://pith.science/paper/ALIXB4WV"},"agent_actions":{"view_html":"https://pith.science/pith/ALIXB4WVUDKIKTEPLQKVN2JEEC","download_json":"https://pith.science/pith/ALIXB4WVUDKIKTEPLQKVN2JEEC.json","view_paper":"https://pith.science/paper/ALIXB4WV","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2402.14146&json=true","fetch_graph":"https://pith.science/api/pith-number/ALIXB4WVUDKIKTEPLQKVN2JEEC/graph.json","fetch_events":"https://pith.science/api/pith-number/ALIXB4WVUDKIKTEPLQKVN2JEEC/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/ALIXB4WVUDKIKTEPLQKVN2JEEC/action/timestamp_anchor","attest_storage":"https://pith.science/pith/ALIXB4WVUDKIKTEPLQKVN2JEEC/action/storage_attestation","attest_author":"https://pith.science/pith/ALIXB4WVUDKIKTEPLQKVN2JEEC/action/author_attestation","sign_citation":"https://pith.science/pith/ALIXB4WVUDKIKTEPLQKVN2JEEC/action/citation_signature","submit_replication":"https://pith.science/pith/ALIXB4WVUDKIKTEPLQKVN2JEEC/action/replication_record"}},"created_at":"2026-07-05T11:39:39.816427+00:00","updated_at":"2026-07-05T11:39:39.816427+00:00"}